Welcome to my AI & Machine Learning portfolio showcase.
This repository aggregates my key projects as Git submodules, making it easy to explore my work in one place.
Each project demonstrates different skills — from fine-tuning LLMs, to building conversational AI interfaces, to deploying Retrieval-Augmented Generation (RAG) systems.
A Retrieval-Augmented Generation system that acts as a personal AI expert on my experience, skills, and projects.
- Context-aware question answering
- Knowledge base powered by my professional background
- Interactive Gradio interface
Exploring regression tasks with Large Language Models - can LLMs predict numerical values effectively?
Key Features:
- Fine-tuned Llama 3.1 with QLoRA optimization
- GPT-4 mini with RAG enhancement
- Comprehensive benchmarking across 9 different models
An autonomous multi-agent AI system that continuously hunts for online deals using intelligent price estimation.
Key Features:
- Multi-agent orchestration (Scanner, Planner, Price Estimator)
- RAG-powered price prediction using ChromaDB and fine-tuned LLM
- Automated RSS feed monitoring from DealNews
- Structured Outputs with Pydantic for reliable data extraction
- Real-time Gradio dashboard with auto-refresh
- Smart filtering (only surfaces deals with $50+ discounts)
A production-ready multi-provider AI chat application with:
- Unified interface for OpenAI, Claude, Gemini, LLaMA, Hugging Face
- Real-time voice integration
- Support for both cloud APIs and local inference (OLLAMA, Hugging Face)
- Streamlit-powered modern UI
- Optimized for performance with async and caching
Since this repository uses Git submodules, clone it with:
git clone --recurse-submodules https://github.com/NEGU93/ML-engineering.gitIf you already cloned without submodules:
git submodule update --init --recursiveJ Agustin Barrachina (NEGU93)
GitHub | LinkedIn | Hugging Face